On Tue, Oct 02, 2018 at 09:59:42AM +0200, Daniel Lezcano wrote:
> On 02/10/2018 07:40, Guo Ren wrote:
> [ ... ]
>
> >>
> >> irq = irq_of_parse_and_map(np, 0);
> >> if (irq <= 0)
> >> return -EINVAL;
> > panic();
> > I want a panic here. Return will make debug confused and directly tell
> > the people where is wrong. It's root-timer for us and it must bootup.
> >
> > If it's a co-timer, I also think return is good.
>
> We don't need a panic in case of failure. If the board as an alternate
> timer it should be able to boot, that is the general rule of thumb for
> these drivers in this directory.
>
> The init timer functions will be called via the timer_probe() which
> browse the timer_of_table (the one with all entries for the
> TIMER_OF_DECLARE macros).
>
> You can see in the code below (from timer_probe.c), there is an error
> message emitted if the driver fails to initialize and another one much
> stronger if no timer was initialize.
>
> This is enough trace IMO.
>
> void __init timer_probe(void)
> {
> struct device_node *np;
> const struct of_device_id *match;
> of_init_fn_1_ret init_func_ret;
> unsigned timers = 0;
> int ret;
>
> for_each_matching_node_and_match(np, __timer_of_table, &match) {
> if (!of_device_is_available(np))
> continue;
>
> init_func_ret = match->data;
>
> ret = init_func_ret(np);
> if (ret) {
> pr_err("Failed to initialize '%pOF': %d\n", np,
> ret);
> continue;
> }
>
> timers++;
Ok, let other timers boot up as possible and no panic for csky timer.
Guo Ren
